2010 jep space experiments

Journal of Experimental Psychology: Applied
2010, Vol. 16, No. 2, 158 –172

© 2010 American Psychological Association
1076-898X/10/$12.00 DOI: 10.1037/a0019292

The Relative Merits of Lean, Enriched, and Empowered Offices:
An Experimental Examination of the Impact of Workspace Management
Strategies on Well-Being and Productivity
Craig Knight and S. Alexander Haslam
University of Exeter
Principles of lean management encourage managers to exert tight control over office space and the people
within it. Alternative, design-led approaches promote the value of offices that are enriched, particularly
by plants and art. On the basis of a social identity perspective, we argue that both of these approaches
may compromise organizational outcomes by disempowering workers and failing to give them input into
the design of their office space. This hypothesis is tested in two experiments (ns ⫽ 112, 47). The first
was conducted in an interior office in a psychology department, the second in a commercial city office.
In 4 independent conditions we examine the impact of space management strategies in which the office
is either (a) lean, (b) decorated by the experimenter (with plants and art), (c) self-decorated, or (d)
self-decorated and then redecorated by the experimenter. We examine the impact of these conditions on
organizational identification, well-being, and various forms of productivity (attention to detail, information processing, information management, and organizational citizenship). In both experiments, superior

outcomes are observed when offices are decorated rather than lean. However, further improvements in
well-being and productivity are observed when workers have input into office decoration. Moreover,
these effects are attenuated if this input is overridden. Implications for theory and practice are discussed.
In particular, findings point to the need to question assumptions about the merits of lean office space
management that have been dominant throughout the last century.
Keywords: identity, space, office, productivity, well-being

Studies of psychological well-being at work were initiated at the
turn of the last century (e.g., Mayo, 1933; Mead, 1913; Myers,
1925; Viteles, 1923; Wells, 1912) and continue to this day (e.g.,
Hansson, Vingard, Arnetz, & Anderzen, 2008; Messer & White,
2006; Mills, Tomkins, & Schlangen, 2007). However, the management of modern office space is typically influenced far less by
psychologists than by architects, interior designers, facility managers, corporate real estate agents, and popular management theorists (Cohen, 2007; Stegmeier, 2008). Here the emphasis is generally on corporate return rather than psychological welfare (Bain
& Taylor, 2000; Handy, 1990). Indeed it has been observed that
when it comes to office management more generally, psychological factors tend to be considered only as an adjunct to business
interests rather than exerting any influence over them (Furnham,
1990; Peters & Waterman, 2004; Statt, 2004).
In this paper we report research that explores some of the key
concepts at the heart of workspace management. In this, we draw
on insights from the social identity approach to organizational life,

as previously applied to the study of office space (e.g., Ashforth &
Mael, 1989; Baldry, Bain, & Taylor, 1998; Haslam, 2004; Knight
& Haslam, in press; Millward, Haslam, & Postmes, 2007; Postmes,
Tanis, & de Wit, 2001). The key issues that we investigated are
whether empowerment within office space impacts on (a) wellbeing (in particular, feelings of psychological comfort, organizational identification, physical comfort, and job satisfaction) and (b)
productivity.

The Lean Approach: The Case for Managerial
Control of Office Space
Key recommendations of the Taylorist approach to office space
management (e.g., Pruijt, 2003; Tapping & Dunn, 2006) include
(a) the removal from the workspace of everything except the
materials required to do the job at hand, (b) tight managerial
control of the workspace, and (c) standardization of managerial
practice and workspace design (Boyer, 2003; Duffy, 1997; C.
Harris & Harris, 2006). These ideas have been particularly influential in work that has promoted the lean office as the key to
efficiency and productivity (Hirano, 1996; Hobson, 2006; Louis,
2007; Tapping & Shuker, 2002; Zalesny & Farace, 1987). This
approach is exemplified by Bibby (1996) in his comparison of two

adjacent offices in a modern bank:

Craig Knight and S. Alexander Haslam, School of Psychology, The
University of Exeter.
This research was supported by a CASE award from the Economic and
Social Research Council (PTA–33–2005– 00021). This award was cosponsored by Ambius (a division of Rentokil Initial) and Haworth U.K. (a
division of Haworth, Inc.).
Correspondence concerning this article should be addressed to Alex
Haslam, School of Psychology, The University of Exeter, Perry Road,
Exeter, EX4 4QG England. E-mail: A.Haslam@ex.ac.uk

The contrast between the old and new in office life is currently well
reflected here. Part of one floor is temporarily being occupied by staff
from the [old] operation: here there is the usual clutter of office
paperwork to be seen, the pinned-up postcards and personal photo158

OFFICE SPACE MANAGEMENT
graphs beside the desks. By contrast, the desks for [new] staff only a
few feet away are spick-and-span, bare of all paper and, in line with
company policy, free of any personal belongings. (para. 10)

The Taylorist literature sees lean, open space as efficient for a
number of reasons. In the first instance, large, uncluttered space
can accommodate more people and so lends itself to economies of
scale (Durmusoglu & Kulak, 2008; Kelliher & Anderson, 2008).
Desks (undecorated or personalized) can also easily be reconfigured for use by other workers (Hobson, 2006; Thompson, 2000).
As a result, space occupancy can be centrally managed with
minimal “disruptive” interference from workers (Keyte & Locher,
2004; Titman, 1991). Indeed, many businesses now adopt a clean
or lean office policy because they have more employees than they
have spaces at which they can work. These lean desks are either
taken on a first-come-first-served basis (hot desking) or can be
booked in advance (hotelling; Millward et al., 2007; Stegmeier,
2008). In the lean office, employee involvement in the running of
the working space is purposefully de-emphasized (Wood & Wall,
2007; Zeisel, 2006). Low-status workers follow the system
planned for them by management (George, Maxey, Rowlands, &
Upton, 2004; Skinner, 2005), performing deskilled, repetitive tasks
(J. A. H. Becker & O’Hair, 2007), reflecting Taylor’s injunction
that “all possible brainwork should be removed from the shop and

centered in the planning or laying out department” (Braverman,
1974, p. 447).
These low autonomy environments echo the demand— control
model (Karasek, 1979), which argues that a combination of lowdecision latitudes and high pressure job roles (e.g., as found in a
classic telesales environment; Baldry et al., 1998), lead not just to
psychologically uncomfortable working conditions and depression
(Seligman, 1975; Sundbom, 1971), but also to greater incidence of
cardio-vascular disease. Indeed since the early 1960s, research has
pointed to the negative association between social class and coronary heart disease (Theorell & Karasek, 1996), despite the persistent belief that those at the top of the tree are under the most
pressure (Martin, 1997; Peters, 1989).
The methodology based on low-worker autonomy has proved
attractive to businesses since Taylor and his contemporaries began
their work in the 1880s (Becker & Steele, 1995; Kanigel, 1999).
Yet despite the enormous body of literature spawned (e.g., Bibby,
1996; Brill, Margulis, Konar, & BOSTI, 1984; George et al., 2004;
Hirano, 1996; Hobson, 2006; Hyer & Wemmerlov, 2002; Louis,
2007; Pruijt, 2003) there is a surprising lack of empirical evidence
to support the claims for greater efficiency. There would appear to
be two main reasons for this oversight. First, the assumption that
Taylorist methodology “just works” (Pyzdek, 2003, p. 664) and,

second, the heavy reliance (particularly in fields of design, architecture, and space management) on evidence gleaned from case
studies (e.g., Louis, 2007; Tapping & Shuker, 2002; Taylor, 1911).

The Green Approach: The Case for Design-Led
Office Space
Space planning and design is frequently seen as an expression of
managerial intent (Marmot & Ely, 2000), in which a building’s
aesthetics are seen as an opportunity to reflect and project a
particular corporate ethos and image (Myerson & Ross, 2003; see
also Cornelissen, Haslam, & Balmer, 2007). We have seen how

159

this space is often deliberately stark (or lean; C. Harris & Harris,
2006; Hobson, 2006), but some organizations choose to avoid
Taylorist prescriptions for a lean office and instead enrich the
workspace by investing in “environmental comfort” (Vischer,
2005, p. 102). This strategy is typically informed by a belief that
such enrichment may promote health. In particular, aesthetically
uplifting art—particularly images from nature—is believed to reduce stress and anger in a working environment (Kweon, Ulrich,

Walker, & Tassinary, 2008). The presence of living plants in a
workspace is also thought to have the additional benefit of cleaning, or “conditioning” the air, thereby helping workers feel happier
and healthier (Bringslimark, Hartig, & Patil, 2007; Dravigne,
Waliczek, Lineberger, & Zajicek, 2008).
In line with these ideas, psychological literature has suggested
that relative to lean offices, enriched offices are psychologically
advantageous (Elsbach, 2003; Handy, 1990; Haslam & Knight,
2006; Myerson, 2007; Thompson, 2000; Vischer, 2005; Zelinsky,
2006). More specifically, it leads to the hypothesis that enriching
workers’ office space with pictures and plants is likely to increase
organizational identification (Hypothesis 1a; H1a), well-being
(specifically, sense of psychological comfort, job satisfaction, and
physical comfort; H2b), and productivity (H2c). In line with previous work (see Knight & Haslam, in press), we would also
anticipate that (as with H2 and H3 below) organizational identification might also mediate the relationship between the experimental condition and other dependent variables.

The Social Identity Approach to Space Use
Yet despite being more sensitive to employees’ needs than lean
approaches, it remains true that even the most benign, designfocused space management strategies still tend to assume that it is
management’s prerogative to retain control of the workspace
(Laing, Duffy, Jaunzens, & Willis, 1998; Peters & Waterman,

2004). This assumption is one that is increasingly being called into
question—not least by designers themselves (Commission for Architecture and the Built Environment, 2004; Froggett, 2001;
Zeisel, 2006). In particular, some psychologists have argued that
employees should be encouraged to decorate their immediate
space with meaningful artifacts to project their identity onto their
own environment and to give some sense of permanency, control,
and privacy (Baldry, 1997; Hall, 1968; Vischer, 2005). It has been
noted that approximately 70% of American workers personalize
their workspaces, but that managers and employees with enclosed
offices decorate more than others (Wells & Thelen, 2002). Where
open-plan offices are common, personalization of low-status working space is both infrequent and discouraged (Laing et al., 1998;
Tapping & Shuker, 2002). Accordingly, it would seem that the
decorative style of one’s working space is primarily predicted by
status (Elsbach & Bechky, 2007; Wells & Thelen, 2002).
At a group level it is argued that collectively, teams should be
free to express their own identity within their workspace, differentiating themselves from other groups without necessarily compromising identification with the organization as a whole (Abrams,
Ando, & Hinkle, 1998; Peters & Waterman, 2004). In particular,
this recommendation is informed by a social identity approach to
organizational life (after Tajfel & Turner, 1979; Turner, Hogg,
Oakes, Reicher, & Wetherell, 1987; Turner, Oakes, Haslam, &

McGarty, 1994), which suggests that employee recognition and

KNIGHT AND HASLAM

160

involvement have the capacity to increase motivation and engagement by increasing organizational identification (i.e., individuals’
willingness to define themselves as members of a particular organizational unit; Ashforth & Mael, 1989; Haslam, Postmes, &
Ellemers, 2003; Tyler & Blader, 2000). In line with this view,
previous research has shown that higher levels of organizational
identification are associated with an increased sense of job satisfaction (Kreiner & Ashforth, 2004; van Dick, 2004) and also with
enhanced group performance (Worchel, Rothgerber, Day, Hart, &
Butemeyer, 1998). Along these lines, a social identity approach to
space management suggests that managers who involve employees
in decision making are also likely to build a sense of shared
identity that enhances the motivation and commitment of junior
colleagues (Ashforth & Mael, 1989; Cornelissen et al., 2007;
Ellemers, De Gilder, & Haslam, 2004). This view is also supported
by Riketta’s (2004) observation of a strong correlation between
shared organizational identification and employees’ willingness to

engage in behavior that is beneficial to both the organization and
job involvement (see also Organ, 1988; van Knippenberg, 2000).
Where decision making is not shared, management is likely to
foster less intrinsic motivation and compliance may be contingent on
higher levels of control and surveillance (Ellemers, van Rijswijk,
Bruins, & de Gilder, 1998; McCabe & Black, 1997; Turner, 1991).
This in turn may lead to lower morale (Ellemers et al., 2004;
Oldham, Hackman, & Pearce, 1976), less cooperative behavior
(Baldry et al., 1998; Organ, 1988; Paille, 2008; Tyler & Blader,
2000) and lower levels of productivity (Vischer, 2005). In this
way, it may also compromise a company’s bottom line (Ellemers
et al., 2004; Hackman & Oldham, 1980; Lawler, 1986). A potential
exception to this rule was highlighted by Ullrich, Christ, and van
Dick (2009) who found that having a leader who is representative
of the group can offset the effects of group members’ lack of
voice—particularly if those members identify highly with the
organizational unit in question. However, these researchers also
found that when leaders were not representative of group members, lack of voice was again associated with negative organizational outcomes.
On the basis of these approaches, we therefore hypothesize that
empowering workers to manage and have input into the design of

their own workspace—thereby allowing them to projecting their
own identity onto it—will enhance organizational identification
(H2a), well-being (H2b), and productivity (H2c).

Reestablishing Managerial Control
Historically, management has not empowered low-status workers (Hobsbawm, 1969; McCabe & Black, 1997). Indeed, the management literature generally counsels that managers should assert
(or reassert) control of the workspace (Pruijt, 2003; Taylor, 1911).
Giving autonomy to workers, only to remove it because management prefers its own options to those chosen by workers is seen by
some literature as a legitimate option (Pruijt, 2003; Tapping &
Shuker, 2002). However, the social identity approach outlined
above would suggest that reintroducing managerial control into
areas where workers are used to more autonomous conditions is
likely to compromise organizational identification and thereby
undermine productivity and well-being (Peters, 1989; Peters &
Waterman, 2004). Along these lines, disempowerment within the
workspace (Frederickson, 1989; George et al., 2004; Titman,

1991) has been found to engender a sense of alienation and
discomfort (Baldry et al., 1998; Handy, 1990) and to reduce job
satisfaction (Ashforth & Mael, 1989; Cohen, 2007). Meanwhile,
research in both environmental design and psychology points to a
link between a reduction in workplace autonomy and greater levels
of stress-related complaint (Bringslimark et al., 2007; Danielsson
& Bodin (2008); Scheepers & Ellemers, 2005). Similarly, a metaanalysis by Humphrey, Nahrgang, and Morgeson (2007) suggests
that an integrated approach that accounts for social needs at work
increases motivation and satisfaction. On the basis of these arguments (Keyte & Locher, 2004; Louis, 2007; Pruijt, 2003; Wood &
Wall, 2007), we therefore predict that disempowering workers by
overriding their input into workspace design will compromise
organizational identification (H3a), well-being (H3b), and productivity (H3c) relative to an enriched or an empowered office environment.

The Present Research
To test the above hypotheses we conducted two experiments in
which space management was manipulated across four independent conditions. In these, the lean condition is informed by a
neo-Taylorist perspective, in which minimalist office space is
intended to focus employees’ attention solely on the work at hand
(in particular through the imposition of a clean desk policy; Bibby,
1996; Fredrickson, 1989; George et al., 2004). The second, enriched condition, instantiates ideas from the design literature in
which workers fulfill their job function in an office that incorporates art and plants, but where they have no input into their
deployment (e.g., Duffy, 1997; Greenhalgh, 2002; Myerson,
2007). A third empowered condition is informed by social identity
principles and allows participants to design their own office environment using a selection of the same art and plants as in the
enriched condition but thereby allowing them to realize something
of their own identity within their working space (De Croon,
Sluiter, Kuijer, & Frings-Dresen, 2005; Elsbach, 2003; Elsbach &
Bechky, 2007). Finally, in a disempowered space, participants’
workspace design is overridden by the experimenter, so that an
initial sense of autonomy within the workspace is taken away
(B. E. Becker & Huselid, 1998; Wood & Wall, 2007).

Experiment 1
In our first experiment, participants were drawn from a wide
cross-section of society and were recruited to take part in a study
that was conducted in a university psychology department. Here
participants were randomly assigned to one of the four experimental conditions described above to gauge the impact of various
space management strategies on organizational identification,
well-being, and productivity with reference to our three main
hypotheses.

Method
Participants and design. There were 112 people (40 men, 72
women) ranging in age from 18 to 78 years (M ⫽ 37.55, SD ⫽
15.05) who took part in the study. Of the sample, 31% described
themselves as students, 61% as being in paid employment, and 8%
as retired. Potential participants were recruited from a range of

OFFICE SPACE MANAGEMENT

sources, but most were drawn from a panel of members of the
general community who had indicated a willingness to participate
in psychological research. Participation was voluntary and unpaid,
although where appropriate, traveling expenses were reimbursed.
Individual participants were randomly assigned to one of four
conditions (lean, enriched, empowered, or disempowered). The
main dependent variables were psychological comfort, organizational identification, job satisfaction, physical comfort, and productivity.
Materials and procedure. The laboratory “office” was a small
interior office in a psychology department, measuring 3.5 m ⫻ 2 m.
The office had no windows or natural light. Participants arrived
individually and it was explained to them that they would take part
in an experiment examining performance on analytical, processing, and intellectual tasks. Participants gave their informed consent
and confidentiality and anonymity were assured.
At every trial, the experimenter (Craig Knight) explained that he
needed to confirm a room booking with a secretary, thus leaving
the participant alone in the office space for 5 min to take in the
ambient environment. The office contained a rectangular desk
(1,600 mm ⫻ 800 mm) and a comfortable office chair on castors.
The room was lit by diffused, overhead fluorescent tubes, the floor
was carpeted and an air conditioning system kept the room at a
constant temperature of 21 °C.
In the lean condition, no further additions to the room were
made. In the enriched condition, participants were shown into a
space where six potted plants (each approximately 350 mm high)
had already been placed toward the edge of the desk surface, so as
not to impinge on the participants’ working area. Six pictures (800
mm ⫻ 800 mm) hung around the walls. The pictures were all
photographs of plants enlarged onto canvas.
In the empowered condition, participants entered an office
where the pictures and plants were placed randomly around the
room. They were told that they could decorate the space to their
taste using as many, or as few, of the plants and pictures provided
as they wished. They could therefore work in a lean or very
enriched space or at a point anywhere along that continuum. The
disempowered condition involved the same initial procedure as the
empowered condition. However, when the experimenter reentered
the office, he looked at the chosen decorations, briefly thanked the
participant and then completely rearranged the pictures and
plants—thereby overriding the participant’s choices. If challenged,
participants were told that their designs were not in line with those
required by the experiment. No further information was given until
the final debrief.
Measures.
Card-sorting task. Once the experimenter returned to the office (or as soon as he had rearranged the pictures and plants in the
disempowered condition), he asked the participant to perform a
card-sorting task. Three packs of playing cards had been shuffled
together and the participant was required to sort them back into the
three constituent packs and to sort each pack into its four suits
(hearts, clubs, diamonds, and spades). These suits then had to be
ordered from ace to king and placed in discreet piles, leaving 12
piles altogether. The key performance measures were the time
taken to complete this task and the number of errors made.
Vigilance task. After this, participants performed a vigilance
task. For this purpose they were given an A4 photocopy of the
same magazine article and asked to cross out and count all the

161

lower case letters “b” that were on the page. The time taken to
complete the task was measured as well as the number of errors
(missed “b”s).
In both cases the participants were told that they needed to
perform the tasks as quickly and as accurately as possible.
Questionnaire. After they had finished both tasks, participants
completed a 74-item questionnaire, in which items measuring
different constructs were presented on five different pages. Most of
these required a response on a 7-point scale ranging from 1
(completely disagree) to 7 (completely agree). The penultimate
page obtained participants’ demographic information. The first
items constituted manipulation checks in which participants were
asked to consider the managerial control of space (Tapping &
Shuker, 2002; Taylor, 1911). There were three, 3-item scales that
examined (a) involvement (␣ ⫽ .87; e.g., “I felt engaged in what
I was doing in the office”; after Lodahl & Kejner, 1965); (b)
autonomy (␣ ⫽ .82; e.g., “During this experiment I had control
over my environment”; after Breaugh, 1989) and (c) quality of
workspace (␣ ⫽ .87; e.g., “This was a pleasant room in which to
work”; after Ferguson & Weisman, 1986).
The scales that followed were all based on previous studies of
space management and organizational identification at work
(Knight & Haslam, in press). Psychological comfort was measured
using a 5-item scale (␣ ⫽ .87; e.g., “I felt at ease during the
experiment”; after Vischer, 2005). Organizational identification
was measured by three items that assessed participants’ identification with the university in which the study was conducted (␣ ⫽
.70; e.g., “I identify with the university”; after Doosje, Ellemers, &
Spears, 1995). Employees’ positive experience of work was assessed using two scales (a) job satisfaction (5 items; ␣ ⫽ .68; e.g.,
“I enjoyed the “finding the letters” task”; after Haslam, O’Brien,
Jetten, Vormedal, & Penna., 2005) and (b) physical comfort (5
items; ␣ ⫽ .75; e.g., “I felt too hot in the room”; after Spector et
al., 2005). After completing the questionnaire, participants were
debriefed and thanked for their participation.

Results
Analytic strategy. Questionnaire and performance data were
analyzed by means of analysis of variance (ANOVA) with office
condition (lean, enriched, empowered, disempowered) as a
between-participants factor. Interitem correlations are shown in
Table 1; means and effect sizes (␩2) are presented in Tables 2 and 3.
Effect sizes indicate how much of the variance in the DV (dependent variable) can be accounted for by each IV (independent
variable). Here an eta-square of .07 is considered moderate, .14
large (Sheshkin, 2004).
Manipulation checks. ANOVAs revealed effects for involvement, autonomy and quality of workspace, Fs(3, 108) ⫽
44.92, 38.21, 20.23, respectively, all ps ⬍ .001, ␩2s ⫽ .56, .51,
.36, respectively. Orthogonal planned contrasts showed that (a)
participants in the lean condition felt less involved, less autonomous, and thought they were in a poorer quality space than
participants in other conditions; (b) that participants in the
enriched office felt less involved and less autonomous than
participants in the empowered condition; and (c) that participants in the disempowered condition felt less involved, less
autonomous, and thought they were in a poorer quality space

KNIGHT AND HASLAM

162

Table 1
Experiment 1: Bivariate Correlations
Dependent variable

1

1.
2.
3.
4.
5.
6.
7.

Involvement
Autonomy
Quality of workspace
Psychological comfort
Organizational identification
Job satisfaction
Physical comfort

ⴱⴱ

p ⬍ .01.

—

2

3
ⴱⴱ

.16
—

than participants in the enriched and empowered conditions.
Relevant statistics are presented in Table 2.
Organizational identification. Analysis revealed a main effect
for organizational identification, F(3, 108) ⫽ 2.87, p ⫽ .04, ␩2 ⫽ .07.
However, orthogonal contrasts only provided support for H3a in
showing that participants in the disempowered condition identified
less with the organization than participants in enriched and empowered conditions. Relevant statistics are presented in Table 2.
Well-being. Analysis revealed effects for psychological comfort,
job satisfaction, and physical comfort, Fs(3, 108) ⫽ 21.15, 5.55,
10.03, respectively, ps ⫽ .001, .001, .001, respectively, ␩2s ⫽ .37,
.13, .22, respectively. Consistent with H1b, orthogonal contrasts
showed that participants in the lean condition felt less psychologically
comfortable, reported less job satisfaction, and expressed lower feelings of physical comfort than participants in other conditions. Consistent with H2b, orthogonal contrasts showed that participants in the
empowered condition felt more psychologically comfortable and reported greater job satisfaction than participants in the enriched condition. Consistent with H3b, orthogonal contrasts showed that participants in the disempowered condition felt less psychologically
comfortable and reported lower feelings of physical comfort than

.65
.01
—

4

5

ⴱⴱ

.46
.02
.75ⴱⴱ
—

.08
.12
.04
.14
—

6

7
ⴱⴱ

.44ⴱⴱ
.07
.57ⴱⴱ
.70ⴱⴱ
.15
.35ⴱⴱ
—

.41
.06
.39ⴱⴱ
.40ⴱⴱ
.30ⴱⴱ
—

participants in the enriched and empowered conditions. Again, relevant statistics are presented in Table 2.
Productivity. Analysis revealed effects for time taken to complete the card-sorting and the vigilance tasks, Fs(3, 108) ⫽ 10.07,
4.44, respectively, both ps ⬍ .01, ␩2s ⫽ .22, .11, respectively.
However, there were no effects for the number of errors made on
either task, Fs(3, 108) ⫽ 1.67, 0.91, ps ⫽ .18, .44, ␩2s ⫽ .04, .02,
respectively. Consistent with H1c, orthogonal contrasts showed that
participants in the lean condition took longer to complete both timed
tasks than participants in other conditions. Consistent with H2c,
orthogonal contrasts showed that participants in the empowered condition took less time to complete the card-sorting task than participants in the enriched condition. Consistent with H3c, orthogonal
contrasts showed that participants in the disempowered condition took
more time to complete the vigilance task than those in enriched and
empowered conditions. Relevant statistics are presented in Table 3.

Discussion
This experiment provided support for our three core hypotheses.
Consistent with H1, relative to the lean condition, participants in

Table 2
Experiment 1: Scale Properties, Means, and Effects for Measures of Participants’ Subjective Experience
Condition (n ⫽ 112)

Effects

Contrasts (t values)

Dependent variable

Lean

Enriched

Empowered

Disempowered

F(3, 108)

Effect size
(full ␩2)

LvR, P,
D, H1a

RvP, H2a

DvR, P, H3a

Involvementb
SD
Autonomyb
SD
Quality of workspaceb
SD
Psychological comfortb
SD
Organizational identificationb
SD
Job satisfactionb
SD
Physical comfortb
SD

2.56
1.06
2.90
1.20
3.32
1.12
4.01
1.11
4.60
1.19
4.82
0.82
4.56
1.29

3.21
1.34
3.77
1.31
5.39
1.09
4.74
0.99
5.25
1.13
5.26
0.93
5.49
0.88

5.77
0.91
5.93
0.85
5.49
1.02
5.72
1.02
4.64
1.14
5.71
0.78
5.74
0.96

3.18
1.15
3.95
0.96
4.57
1.06
4.24
1.06
4.33
1.37
5.31
0.74
4.59
0.90

44.92ⴱⴱ

.56

6.08ⴱⴱ

8.50ⴱⴱ

5.04ⴱⴱ

38.21ⴱⴱ

.51

6.90ⴱⴱ

7.37ⴱⴱ

3.55ⴱⴱ

20.23ⴱⴱ

.36

7.10ⴱⴱ

0.32

3.18ⴱⴱ

21.15ⴱⴱ

.37

4.64ⴱⴱ

4.22ⴱⴱ

4.19ⴱⴱ

2.87ⴱ

.07

0.56

1.88

2.17ⴱ

5.55ⴱⴱ

.13

3.39ⴱⴱ

2.08ⴱ

0.91

.22

ⴱⴱ

0.92

4.37ⴱⴱ

10.03

ⴱⴱ

3.19

Note. L ⫽ lean; R ⫽ enriched; P ⫽ empowered; D⫽ disempowered; H1 ⫽ Hypothesis 1, (L ⬍ R, P, D); H2 ⫽ Hypothesis 2, (D ⬍ R); H3 ⫽ Hypothesis
3, (D ⬍ R, P, H).
a
Means relate to 7-point scales (1 [completely disagree]–7 [completely agree]). b Contrast related to relevant hypotheses.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01.

Note. L ⫽ lean; R ⫽ enriched; P ⫽ empowered; D⫽ disempowered; H1 ⫽ Hypothesis 1, (L ⬍ R, P, D); H2 ⫽ Hypothesis 2, (D ⬍ R); H3 ⫽ Hypothesis 3, (D ⬍ R, P, H).
Contrast related to relevant hypotheses. b Time given in minutes. c Number of errors. d Total time and total errors (productivity) are shown as a percentage of the control (Lean) condition.
†
p ⬍ .10. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01.
a

1.23
0.00
3.02ⴱⴱ
1.41
2.73ⴱⴱ
1.39
2.51ⴱ
2.15ⴱ
1.21
0.37
2.95ⴱⴱ
0.30
4.74ⴱⴱ
0.61
1.66†
0.77
4.81ⴱⴱ
0.93
.22
.04
.11
.02
.27
.03
10.07ⴱⴱ
1.67
4.44ⴱⴱ
0.91
13.11ⴱⴱ
0.96
12.76 (2.70)
0.82 (1.22)
7.70 (2.45)
19.82 (5.40)
20.47, 90.0% (4.28)
20.64, 100.3% (5.96)
10.94 (2.64)
0.36 (0.95)
6.08 (1.20)
18.21 (4.89)
16.74, 73.6% (2.39)
18.36, 89.3% (4.97)
15.24 (3.20)
1.04 (2.02)
7.51 (1.92)
19.54 (6.25)
22.75, 100% (4.12)
20.57, 100% (6.80)
Card sorting taskb (SD)
Card sorting taskc (SD)
Vigilance taskb (SD)
Vigilance taskc (SD)
Total timed (SD)
Total errorsd (SD)

12.91 (3.19)
1.29 (1.98)
6.69 (1.76)
17.64 (6.52)
19.60, 86.2% (3.47)
18.86, 91.2% (7.40)

RvP, H2a
LvR, P,
D, H1a
Effect size
(full ␩2)
F(3, 103)
Empowered
Enriched
Lean
Dependent variable

Condition (n ⫽ 112)

Table 3
Experiment 1: Means and Effects for Performance Measures

Disempowered

Effects

Contrasts (t values)

DvR,
P, H3a

OFFICE SPACE MANAGEMENT

163

enriched office space reported enhanced feelings of organizational
identification and well-being, in line with previous claims made in
the design literature (Elsbach & Beckhy, 2007; Zelinsky, 2006). It
also led to the tasks being performed quicker, with no decrement
in accuracy.
When participants were empowered to decorate their own working space, this led to further improvements in participants’ perceptions of their working conditions. Consistent with H2, empowerment within the office space improved feelings of well-being
(Faller, 2002; Haslam, Eggins, & Reynolds, 2003; Postmes et al.,
2001; van Dick, Ullrich, & Tissington, 2006). Tasks were also
completed more quickly but, more important, without any accompanying rise in errors. However, once this feeling of empowerment
was overridden by the experimenters (i.e., in the disempowered
condition), as predicted by H3, feelings of organizational identification and well-being fell relative to those of participants in both
the enriched and the empowered conditions. Disempowerment also
led participants to take longer to complete the two tasks (Peters,
1989).
Despite the support that it provides for our hypotheses, this first
study also has some significant limitations. First, our sample
represented a fairly wide cross-section of the population who had
not necessarily experienced office work themselves. Second, the
experiment took place in a university setting, whereas (for obvious
reasons) the majority of previous design studies have been based in
the workspace (e.g., Brill et al., 1984; Dravigne, Waliczek,
Lineberger & Zajicek, 2008; Gensler, 2005; Gorjup, Valverde, &
Ryan, 2008; Louis, 2007).
Third, although support for our hypotheses was generally
strong, it was noticeably weaker on the measure of organizational
identification. Indeed, the fact that there was no support for H1a or
H2b meant that there could be no evidence of organizational
identification mediating the relationship between the way space
was managed and participants’ well-being and productivity (as
found in the previous survey research; Knight & Haslam, in press).
In part, this may reflect the fact that identification was here
operationalized by asking participants to express their level of
identification with an organizational entity (the university), that
was irrelevant both to their everyday lives and to this study. More
relevant, then, was their identification with those who conducted
the study itself. However, this was something we failed to assess.
Another concern was that the card-sorting task could be seen as
unrepresentative of tasks typically performed in an office environment (Anastasi, 1988). Finally, this study did not include specific
measures of organizational citizenship behavior (OCB; Organ,
1988) that might have allowed us to examine issues of workspace
motivation and consideration. Along these lines, OCB is seen as a
key indicator of relevant outcomes at the organizational level
because it measures “behavior that is discretionary, not directly or
explicitly recognized by the formal reward system, and in the
aggregate promotes the efficient and effective functioning of the
organization” and so would seem to be particularly important to
address in this context (Organ, 1988, p. 4; see also Baker, Hunt, &
Andrews, 2006; Messer & White, 2006; Tyler & Blader, 2000).

Experiment 2
Experiment 1 provided evidence that participants performed and
felt better having been involved in decisions that affected their

164

KNIGHT AND HASLAM

workspace. As hypothesized, nonempowered and disempowered
participants (after Baldry et al., 1998; Laing et al., 1998; Sewell,
1998; van Dick, Christ, & Stellmacher, 2004) were less satisfied
and less productive than participants who were empowered. Nevertheless, as outlined above, the study had significant limitations.
To address these issues, Experiment 2 used a sample drawn exclusively from a population of office workers. The experiment
itself took place in a working office and contained more realistic,
office-based tasks. The study also included a more relevant measure of organizational identification and an explicit measure of
organizational citizenship behavior (Organ, 1988). These tasks
were designed to replicate the straightforward tasks (information
processing and management) and repetitive activities (vigilance)
found in many low-skilled office jobs (C. Harris & Harris, 2006).
Although recognizing that self-reported OCB is of only limited
validity (Baker et al., 2006); it was felt that developing a quantifiable measure of citizenship behavior would usefully augment the
results of the study. On the basis of social identity theorizing, we
anticipated that OCB would increase to the extent that workers
identify with each other and with their employer (Haslam, 2004;
Postmes et al., 2001). Thus, although the hypotheses for Experiment 2 were the same as those in Experiment 1, we also predicted
that OCB would be more apparent in an enriched office than a lean
office (H1d), that OCB would increase further in an empowered
space (H2d), and that it would be reduced if empowered workers
were subsequently disempowered (H3d).

Method
Participants and design. The design of Experiment 2 was the
same as Experiment 1 but with the addition of a quantifiable
measure of OCB. There were 47 office workers (28 men, 19
women) ranging in age from 22 to 61 years (M ⫽ 36.23, SD ⫽
9.57) who took part in the study. Of the sample, 35% described
themselves as nonmanagement staff, 30% as lower management,
26% as middle management, and 9% as senior management.
Potential participants, all from commercial businesses, were contacted by mail, email, and telephone. Participation was on a voluntary basis and was unpaid.
Materials and procedures. The study was conducted in London in an air-conditioned commercial office approximately 4.5 m ⫻
6 m in size. The space housed an executive desk (approximate
dimensions 2,200 ⫻ 800 mm) with two, 90° returns of approximately 2,000 ⫻ 600 mm, so that the effective desktop area took up
three sides of a hollow square with the participant at its center.
There was also a large matching credenza with eye-level storage in
the room (approximately 1,800 high ⫻ 2,200 wide ⫻ 800 deep).
Participants sat in a high backed, comfortable leather chair as they
worked. The room had a raised Tec-Crete floor, with a large
sea-grass rug beneath the desk covering the immediate working
area. The door and most of the walls were glass. Given that the
office used in Experiment 1 had no windows, views and external
distractions in this second experiment were minimized by ensuring
that participants sat with their backs to the outside world. Meanwhile temporary, opaque transfers were fixed to all other areas of
glass up to eye-line height thus obscuring any further views. The
study followed the same procedure as Experiment 1, using the
same number of plants and pictures. Participants spent 5 min alone

in their workspace in which to absorb the ambient environment
before the experiment began.
Measures. Although the instructions and timing procedures
were the same as in Experiment 1, the measures in Experiment 2
varied slightly to include tasks that were more representative of an
office environment (Anastasi, 1988). An additional OCB element
was also added.
Information management and processing task. Participants
were asked to work with a shuffled pile of corporate memoranda
based on a fictitious company. They had to imagine that they were
employees of this company and (a) sort the memoranda into
chronological order (an information management task) and then
(b) answer 15, multiple-choice questions based on the information
contained in these memos (an information processing task).
Vigilance task. The experiment’s second element, was exactly
the same as in Experiment 1 and once more participants were told
that they needed to perform the tasks as quickly and accurately as
possible.
Organizational citizenship behavior task. This new measure
took the form of a quantifiable, OCB task (after Organ, 1988;
Williams Pitre, & Zainuba, 2000). This built on the participants’
fictitious employment with the company described in the information management task. Participants were asked to imagine that in
addition to a normal workload, they were responsible for 10 further
tasks. Five of these were undesirable (e.g., “Draw up proposals
about how the company should reduce its headcount”) and five
were desirable (e.g., “Represent the company at the annual Awards
Dinner”; after Paille, 2008). Participants were told that any number
of these tasks could be off-loaded onto a colleague and that this
would have no additional implications for them as the company’s
management would make sure that the participants’ peers did not
find out the source of any increase in workload.
Questionnaire. The same questionnaire was used as in Experiment 1, but with two modifications. The three-item, organizational identification scale now reflected participants’ identification
with the organization managing (rather than the organization hosting) the experiment (␣ ⫽ .90; typical item: “I identify with the
organization that is running this experiment”; after Doosje et al.,
1995), while the job satisfaction scale incorporated a measure of
OCB (8 items; ␣ ⫽ .90; e.g., “If these were my normal working
conditions I would stay behind to do extra work if necessary, even
if I was not paid overtime”; after Haslam, O’Brien, Jetten,
Vormedal, & Penna, 2005).

Results
Analytic strategy. Questionnaire and performance data were
then analyzed by means of ANOVA with office condition (lean,
enriched, empowered, disempowered) as a between-participants
factor. Interitem correlations are shown in Table 4; means and
effect sizes (␩2) are presented in Tables 5 and 6.
Manipulation checks. ANOVAs revealed effects for involvement, autonomy and quality of workspace, Fs(3, 43) ⫽ 18.42,
29.96, 11.51, respectively, all ps ⬍ .001, ␩2s ⫽ .57, .68, .45,
respectively. Orthogonal planned contrasts indicated that (a) participants in the lean condition felt less involved, less autonomous,
and thought they were in a poorer quality space than participants
in other conditions; (b) participants in the enriched office felt less
involved and less autonomous than participants in the empowered

OFFICE SPACE MANAGEMENT

165

Table 4
Experiment 2: Bivariate Correlations
Dependent variable

1

1.
2.
3.
4.
5.
6.
7.

Involvement
Autonomy
Quality of workspace
Psychological comfort
Organizational identification
Job satisfaction
Physical comfort

ⴱⴱ

p ⬍ .01.

—

2
.01
—

condition; and (c) participants in the disempowered condition felt
less involved, less autonomous, and thought they were in a poorer
quality space than participants in the enriched and empowered
conditions. Relevant statistics are presented in Table 5.
Organizational identification. Analysis revealed a main effect for organizational identification, F(3, 43) ⫽ 4.29, p ⬍ .01,
␩2 ⫽ .23. However, although this effect was stronger than that
obtained in Experiment 1, orthogonal contrasts again only provided support for H3a in showing that participants in the disempowered condition identified less with the organization than participants in enriched and empowered conditions. Relevant statistics
are presented in Table 5.
Well-being. Analysis revealed effects for psychological comfort, job satisfaction, and physical comfort, Fs(3, 43) ⫽ 20.50,
7.00, 6.65, respectively, all ps ⬍ .01, ␩2s ⫽ .65, .33, .32, respectively. Consistent with H1b, orthogonal planned contrasts indicated that participants in the lean condition felt less psychologically comfortable, reported less job satisfaction, and felt less
physically comfortable than participants in other conditions. Consistent with H2b, orthogonal contrasts showed that participants in
the empowered condition felt more psychologically comfortable
than participants in the enriched condition. Consistent with H3b,
orthogonal contrasts indicated that participants in the disempowered condition felt less psychologically comfortable, reported
lower levels of job satisfaction, and reported feeling less physically
comfortable than participants in enriched and empowered conditions. Relevant statistics are again presented in Table 5.

3
ⴱⴱ

.89
.04
—

4

5

ⴱⴱ

6

ⴱⴱ

.72
.12
.79ⴱⴱ
—

7

ⴱⴱ

.68
.07
.62ⴱⴱ
.64ⴱⴱ
—

.55
.05
.55ⴱⴱ
.60ⴱⴱ
.72ⴱⴱ
—

.60ⴱⴱ
.02
.62ⴱⴱ
.61ⴱⴱ
.67ⴱⴱ
.60ⴱⴱ
—

Productivity. Analysis revealed effects for time taken to complete both the information management and the vigilance tasks,
Fs(3, 43) ⫽ 3.73, 5.75, respectively, ps ⫽ .018, .002, respectively,
␩2s ⫽ .21, .29, respectively. It also revealed effects for the number
of errors made on the information management task F(3, 43) ⫽
4.17, p ⫽ .011, ␩2 ⫽ .23. At the same time there were no effects
for the number of errors made on the vigilance task, F(3, 43) ⫽
1.23, p ⫽ .311, ␩2 ⫽ .08. Consistent with H1c, orthogonal planned
contrasts showed that participants in the lean condition took longer
to complete the information management task. There were no
significant differences in the number of errors made on the information management task. Consistent with H2c, orthogonal contrasts showed that participants in the empowered condition took
less time to complete the vigilance task than participants in the
enriched condition. There were no significant differences in terms
of the number of errors made on the information management task.
Consistent with H3c, orthogonal contrasts showed that participants
in the disempowered condition took longer to complete the vigilance task than participants in either the enriched or the empowered conditions. Participants in the disempowered condition also
made significantly more errors on the information management
task than those in the enriched or empowered conditions. Relevant
statistics are presented in Table 6.
Organizational citizenship. Analysis revealed a main effect
for the total number of tasks retained on the OCB task, F(3, 43) ⫽
4.77, p ⫽ .006, ␩2 ⫽ .25. Consistent with H1d, orthogonal planned
contrasts indicated that participants in the lean condition retained

Table 5
Experiment 2: Scale Properties, Means, and Effects for Measures of Participants’ Subjective Experience
Condition (n ⫽ 47)
Dependent variable

Lean

2.25 (1.37)
Involvementb (SD)
2.44 (1.33)
Autonomyb (SD)
3.39 (1.51)
Quality of workspaceb (SD)
3.02 (1.44)
Psychological comfortb (SD)
Organizational identificationb (SD) 4.00 (2.10)
b
3.51 (1.69)
Job satisfaction (SD)
5.02 (1.28)
Physical comfortb (SD)

Enriched
4.82 (0.91)
4.33 (1.22)
5.47 (1.32)
4.98 (1.06)
5.19 (1.19)
5.06 (1.13)
6.08 (1.55)

Effects

Empowered Disempowered F(3, 43)
5.83 (0.66)
5.94 (1.28)
5.69 (1.40)
5.68 (1.23)
5.00 (1.13)
5.25 (0.98)
6.45 (1.25)

2.79 (1.24)
2.48 (1.50)
3.82 (1.44)
3.20 (1.64)
3.21 (1.41)
4.02 (1.75)
4.79 (1.25)

18.42ⴱⴱ
29.96ⴱⴱ
11.51ⴱⴱ
20.50ⴱⴱ
4.29ⴱⴱ
7.00ⴱⴱ
6.65ⴱⴱ

Contrasts (t values)

Effect size LvR, P,
D, H1a RvP, H2a DvR, P, H3a
(full ␩2)
.57
.68
.45
.65
.23
.33
.32

4.93ⴱⴱ
5.14ⴱⴱ
4.08ⴱⴱ
5.45ⴱⴱ
0.93
3.49ⴱⴱ
2.11ⴱ

Note. H1 ⫽ Hypothesis 1, (L ⬍ R, P, D); H2 ⫽ Hypothesis 2, (R ⬍ P); H3 ⫽ Hypothesis 3, (D ⬍ R, P).
a
Contrast related to relevant hypotheses. b Means relate to 7-point scales (1 [completely disagree]–7 [completely agree]).
†
p ⬍ .10. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01.

1.81†
3.75ⴱⴱ
0.46
1.95#
0.32
0.43
0.84

5.14ⴱⴱ
6.92ⴱⴱ
4.12ⴱⴱ
6.66ⴱⴱ
3.43ⴱⴱ
2.87ⴱⴱ
3.81ⴱⴱ

KNIGHT AND HASLAM
Note. H1 ⫽ Hypothesis 1, (L ⬍ R, P, D); H2 ⫽ Hypothesis 2, (R ⬍ P); H3 ⫽ Hypothesis 3, (D ⬍ R, P); OCB ⫽ organizational citizenship behavior.
Contrast related to relevant hypotheses. b Time given in minutes. c Given in errors. d Total time and total errors (productivity), and number of tasks retained (OCB) are shown as a percentage
of the control (Lean) condition.
†
p ⬍ .10. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01.
a

1.33
3.52ⴱⴱ
3.42ⴱⴱ
1.07
1.98†
1.70†
1.67
0.24
1.13
1.29
0.44
2.24ⴱ
1.01
1.67
0.95
0.88
1.18
0.40
.21
.23
.29
.08
.24
.12
.23
.08
.25
25.49 (6.85)
1.00 (1.21)
8.03 (2.34)
20.00 (9.97)
33.53, 82.9% (7.62)
20.42, 83.9% (8.91)
3.42 (1.17)
3.25 (.87)
6.42, 142.7% (1.78)
32.04 (9.28)
1.42 (1.08)
8.42 (2.11)
22.92 (10.16)
40.45, 100% (10.26)
24.33, 100% (10.02)
2.00 (1.91)
2.50 (1.31)
4.50, 100% (1.98)
Information management/handling
taskb (SD)
Information managementc (SD)
Vigilance taskb (SD)
Vigilancec (SD)
Total timed (SD)
Total errorsd (SD)
Negative OCB tasks retained (SD)
Positive OCB tasks retained (SD)
Total OCB tasks retainedd (SD)

21.29 (6.63)
0.75 (.97)
6.13 (1.78)
16.33 (8.99)
27.41, 67.8% (6.74)
17.08, 70.2% (6.52)
3.92 (.79)
2.75 (.97)
6.67, 148.2% (1.24)

27.27 (8.91)
2.64 (2.06)
9.67 (2.06)
21.64 (9.98)
36.94, 91.3% (9.70)
24.09, 99.0% (9.87)
2.82 (1.42)
3.09 (.94)
5.91, 131.3% (1.30)

3.73ⴱ
4.17ⴱ
5.75ⴱⴱ
1.23
4.58ⴱⴱ
1.88
4.20ⴱ
1.25
4.77ⴱⴱ

2.75ⴱ
0.10
0.68
1.21
2.61ⴱ
1.32
2.97ⴱⴱ
1.53
3.56ⴱⴱ

DvR, P,
H3a
RvP, H2a
LvR, P,
D, H1a
Effect size
(full ␩2)
F(3, 43)
Empowered
Enriched
Lean
Dependent variable

Table 6
Experiment 2: Means and Effects for Performance Measures

Condition (n ⫽ 47)

Disempowered

Effects

Contrasts (t-values)

166

fewer OCB tasks (particularly, fewer negative tasks) than participants in other conditions. No other contrasts were significant (i.e.,
there was no support for H2d or H3d). Relevant statistics are
presented in Table 6.

Discussion
The findings from this experiment are consistent with those
from Experiment 1 and provide further support for our experimental hypotheses. Consistent with H1, relative to the lean condition,
enriched office space led to improved feelings of psychological
comfort, job satisfaction, and physical co